Reusing Security Video for Customer Analytics

In retail settings, video analytics often requires expensive camera setups and hours of recording. But it doesn't have to. Most retail outlets already have thousands of hours of footage in their security camera banks. And even if they erase the recordings regularly, a week or a month is plenty to get started with.

Startup Prayas Analytics, a company run by two juniors in the Wharton School at University of Pennsylvania, just signed its first giant deal with a top-50 retailer. It will apply its software to security recordings from 1,200 national stores and analyze it to improve stock layout and reduce customer downtime.

The company focuses on two major things, improving shopping efficiency by helping customers find what they want faster, and reducing wait times at the checkout by analyzing cashier speed.

Prayas's software looks at what customers are buying and where in the store they go to get it. If popular items are near the rear of the store, the report may indicate that retailers should shift merchandise around to reduce wasted foot time. Likewise, unpopular products may be moved to more prominent spots or to less conspicuous areas to avoid clutter.

In an interview with their school news site, Prayas founders Yash Kothari and Pranshu Maheshwari said they're already looking to expand into other sectors, including healthcare, which they feel could be made more efficient with analysis of staff routines and the routes that they take through the facility.

Shortening long routes and cutting back on bottlenecks during busy hours can make any business more efficient, and the video footage is already on file.

Who's watching whom?
The pair acknowledged that there are some privacy concerns when it comes to pre-recorded footage, especially when it's handed off to a private firm like theirs. However, they stressed in the Daily Pennsylvanian that their focus is on the store and the masses of customers, not individual people.

Most of the time, security recordings aren't detailed enough to pick out faces anyway, but either way, Prayas Analytics doesn't look at customer-specific factors right now.

The boys will need to stay on their toes, though. They may have the edge on a lot of analytics companies now, but it won't be long before others pick up on this trend and begin offering a comparable service. In fact, Prism Skylabs is already offering something very similar, though it focuses more on immediate and real-time tracking of customers.

Through its web-based app, Prism lets its clients view customers (blurred in-the case of high-def cameras) shopping in real time, check their dwell time on individual products, see the directions people take when browsing the aisles, and even receive text updates when certain targets are met.

On top of that, Prism offers to store all of that data -- including the original feed -- in the cloud.

Millions and millions of hours of security footage are recorded every day by corporations around the US and the rest of the world, and it's just waiting for analytics to unleash its true value.

Jon Martindale, Technology Journalist

Jon Martindale is a technology journalist and hardware reviewer, having been covering new developments in the field for most of his professional career. In that time he's tested the latest and greatest releases from the big hardware companies of the world, as well as writing about new software releases, industry movements,and Internet activism.

Brexit Negotiations Drive Analytics Growth

Could Britain's exit from the EU drive a new wave of analytics investment and growth? Here's a closer look.

Vocal Commands Arrive for Analytics

Voice interfaces may give many more users access to self-service analytics. Here's a closer look.

Keep it simple, silly
  • 6/9/2014 5:44:31 PM

I'm a little dubious about the use of security videos for understanding customers. What about that good old-fashioned approach of actually talking to people/shoppers to get real insights and discern those with real intention from the impulse shopper?

Hardware retailer Lowe's does a really smart thing when it moves into a location where there's already a Home Depot: Lowe's lays out its stores just like Home Depot does, so if you know where the flooring and carpeting is in one, you can easily find it in the other. Brilliant, and ridiculously simple.


Re: Good idea but I can see some issues
  • 6/9/2014 7:09:42 AM

I'm curious, do you have numbers that show how many retailers have this kind of camera coverage? I don't doubt that very large retailers have fairly comprehensive coverage but I've seen enough small retailers who only catch specific areas like the door and cash register that it makes me wonder how big the market would be if you only focused on those that had 90% coverage or better.

Re: Good idea but I can see some issues
  • 6/6/2014 5:18:54 PM

It's an interesting idea that video footage from a few years ago could possilby be analyize customer behavior, even though it may not be what it was orginally intended for.

Re: Good idea but I can see some issues
  • 6/6/2014 12:09:46 PM

We talked about a similar idea in this video: Keeping a Close Eye on Customers. Intel was showing a solution that used security footage from the store entrance to estimate the number of shoppers in the store and staff the checkout lanes appropriately. This is a little more elaborate, and it assumes that there are cameras placed throughout the store. 

In my experience, many stores have cameras covering almost every inch of the space, because they're concerned not only with customer malfeasance, but with employee misbehavior, too.

Re: What to Optimize?
  • 6/6/2014 11:39:45 AM

There are items that draw people to the stores.  And there are other items (batteries for example) that we don't think about much until we see them in the store and think "Oh, I'd better pick up some AA batteries"

If the store aims to minimize the shoppers time, sales of accessories may drop sharply.

I wouldn't suggest a layout that wastes customer's time - things need to be easy to find. But there needs to be a balance.

Re: What to Optimize?
  • 6/6/2014 10:37:23 AM

I'm sure it depends on what the data shows, PC. If those items aren't flying off the shelves because of their placement, the store can experiment with different layouts for an optimal mix. 

Good idea but I can see some issues
  • 6/6/2014 7:46:17 AM

I can see where some of the footage can be shared and how it may be handy for customer analytics but I think they may run into some issues with camera positioning.   From the standpoint of security cameras are typically placed in areas that are hard for a human employee so see from their regular work station or they cover the areas most likely to be involved in a theft.  That means that coverage in many retail stores will be lacking as you get into the areas that management and security cares about a little less.  I can see this solution being a good start though and a way to add value to an existing camera system.

What to Optimize?
  • 6/5/2014 5:50:30 PM

If popular items are near the rear of the store, the report may indicate that retailers should shift merchandise around to reduce wasted foot time.

All this time, I thought the milk and bread were always at the back of the store, intentionally.  It's so you have to walk by the various end-aisle displays.  Maybe you'll remember that you need something else while you're walking through the store.

Is this different for department stores?